A fraud management system architecture for next-generation networks.

Area of Science:

• Telecommunications Engineering
• Network Security
• Artificial Intelligence

Background:

• Traditional fraud management systems (FMS) struggle with the flexibility and application-independent requirements of convergent Next-Generation Networks (NGNs).
• Internet protocol (IP)-based networks generate vast amounts of data, necessitating advanced detection methods.
• Emerging IP-based billing standards like IP Detail Record (IPDR) offer new avenues for fraud analysis.

Purpose of the Study:

• To propose an original, flexible, and application-independent fraud management system (FMS) architecture for Next-Generation Networks (NGNs).
• To leverage artificial intelligence, specifically self-organizing maps (SOMs), for enhanced fraud detection in IP-based networks.
• To demonstrate the effectiveness of the proposed architecture and SOMs through a prototype implementation.

Main Methods:

• Development of a novel four-stage fraud detection architecture for NGNs.
• Analysis of billing records in IP Detail Record (IPDR) format.
• Implementation and testing of a prototype utilizing self-organizing maps (SOMs) for fraud detection.

Main Results:

• The proposed architecture offers superior flexibility and application-independency compared to traditional FMS.
• Self-organizing maps (SOMs) proved effective in uncovering previously unknown NGN fraud scenarios.
• Prototype testing validated the feasibility and effectiveness of the SOM-based approach for fraud detection.

Conclusions:

• The novel FMS architecture is well-suited for the dynamic environment of IP-based NGNs.
• The integration of SOMs represents a significant advancement in detecting sophisticated and emergent fraud patterns.
• The developed prototype confirms the practical applicability of this AI-driven fraud management strategy.